Smart Prosthetic Leg Uses AI to Make Perfect Steps

By | November 1, 2019

Powered prosthetic devices tend to be heavy and not very smart at interpreting the needs of their users. Leg prostheses, for example, have to bend their joints to match whatever the person wants to do, be it sitting down, standing up, or walking up a flight of stairs. Moreover, to feel natural and provide fluid and efficient movement, a powered prosthetic leg has to time the activation of its motors just right at every step.

Researchers at the University of Utah have now built a prototype leg prosthesis that can anticipate and respond to the actions of its users, providing a very natural walking gait, improving the overall balance, and walking power.

The device relies on bespoke force and torque sensors, in addition to off-the-shelf gyroscopes and accelerometers, to know what the leg is doing at all times. An on-board computer evaluates these data, using artificial intelligence algorithms, to identify what the person is doing with the leg and what kind of situation they’re in; whether they’re trying to get up from a chair or taking another step up the stairs, for example.

Based on this, the motors powering the ankle and knee joints are activated to match the intentions of the user and to continually provide optimal balance.

The leg consists primarily of aluminum and titanium, light metals that allow the entire device to weigh in at about 6 pounds (2.7 kg). Small batteries also help to bring the weight down.

In early tests so far, the leg is showing impressive performance, letting users walk with less irritation to their stump and giving them a walking gait they haven’t experienced with other prostheses. Moreover, some of the study participants are elderly, and the new prosthetic allows them to walk in a way they wouldn’t be able to again, since typical powered prostheses are generally designed for younger, stronger people.

Here’s a University of Utah video featuring a Santa Claus outfitted with the prototype leg prosthesis:

Via: University of Utah